What drives political support? Evidence from a survey experiment at the onset of the corona crisis

ABSTRACT

In times of severe crises, citizens are frequently found to rally-‘round-the-flag – i.e. to increase trust in their government. Drawing on an original survey experiment with real-world information at the onset of the coronavirus pandemic, we portray the saliency of rally-relevant issues of the crisis. Contrary to our expectation, this priming effect lowers governmental trust. Mediation analyses show that respondents are more trusting towards their in-group, but simultaneously display less pride in institutions and less positive attitudes towards the political elite, which leads to a reduction in governmental trust. Our study shows that crises do not always lead to a rally-‘round-the-flag effect – it crucially depends on whether people consider their democratic institutions capable of coping with the crisis. Moreover, our findings demonstrate the need to unravel the black box of the rally-effect to get a more accurate picture of the driving forces behind governmental trust during crises.

KEYWORDS:

• Political trust
• Rally-‘round-the-flag
• Coronavirus pandemic
• Priming

Data availability statement

The data that support the findings of this study are openly available in the OSF data repository at https://doi.org/10.17605/OSF.IO/C6VUB.

Acknowledgments

An earlier version of this paper was presented at the Political Trust in Crisis Conference, a research workshop at the University of Bern and the Free University of Berlin as well as a research seminar at the University of Bern. We are grateful to Rahel Freiburghaus and the other participants, Julian Michel as well as the two anonymous referees and the editor for their valuable feedback and suggestions, which have helped improve this paper tremendously.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethics declaration

The research design was approved by the Research Ethics Committee at the Faculty of Business, Economics and Social Sciences of the University of Bern (approval number 092020).

Notes

1 Nevertheless, Amat et al. (2020) also show that exposure to the coronavirus in the personal network may instead decrease political support.

2 Together with Sweden, Switzerland is often regarded as taking a more lenient approach to containing Covid-19. Unlike Sweden, however, Switzerland did not adopt this less stringent approach until Summer 2020. At the time we conducted our survey experiment during the onset of the pandemic, government response stringency in Switzerland was similar to its neighbor countries.

3 Since our survey covered different facets of the coronavirus crisis, the full experiment included four different treatment conditions and one control condition. The conditions were randomized and are thus fully independent of each other. In this paper, we focus on the condition treating mechanisms related to the rally effect.

4 The position of head of government rotates yearly between the seven members of the government (Federal Council) in order of seniority.

5 It is taken from reports shown in the SRF (Swiss Radio and Television). Moreover, the statements of the head of government are part of an open letter to Swiss citizens, which was disseminated by many different news agencies. In this vein, this reflects typical information that citizens encounter when reading the newspapers or watching TV, which is why we assume that our findings can at least partly be applied to real-world relationships outside of our experimental set-up.

6 Our treatment includes both mechanisms simultaneously. We do not attempt to isolate these effects, which is known as a compound treatment (Hernán & VanderWeele, 2011). Rally literature also argues that this two mechanisms are not mutually exclusive, but reinforce each other (Hetherington & Nelson, 2003; Murray, 2017).

7 Information on the randomization of the experimental groups and a balance test can be found in the online appendix in Table A1, Table A2 and Table A3. Out of the 31 t-tests, none were significant at the 10-% level. Overall, we conclude that randomization has worked as intended.

8 Rally literature has employed a wide range of measures to assess public support for the government, including approval ratings of incumbent parties or leaders (Baker & Oneal, 2001), vote intentions (Lai & Reiter, 2005) or vote shares (Bechtel & Hainmueller, 2011) for the incumbent parties as well as trust in the government (Dinesen & Meier Jaeger, 2013; Hetherington & Nelson, 2003). Due to the consensus government nature of the Swiss Federal Council, we opted against measures focusing on the incumbent parties or leaders and instead use the more general trust in government measure.

9 As can be seen in Table 2, model (2) and Table A5, panel (a) in the online appendix, this does not affect our results.

10 These models control for socio-demographics such as age (squared), sex, education, the type of community people live in and self-reported health, an important factor in times of a pandemic, as well as external events of the crisis through a series of dummy variables for the day the respondents filled out the survey. As can be seen in Table 2, model (3) and Table A5, panel (b) in the online appendix, this does not affect our results.

11 Due to the low number of respondents (990 overall, only 316 of which have received the experimental treatment), we also interpret results at the 10% significance level.

12 As a robustness check, we also tested whether the effect differs between respondents on different sides of the political spectrum, since the political affiliation of the head of government may affect how the message is perceived. Given the nature of Swiss governments as consensus governments with a high degree of collegiality, this is unlikely, however, and can be empirically refuted.

13 In this regard, literature has observed that the strength of the rally effect also depends on the severity of a crisis (e.g., Schraff, 2021). To rule out this alternative mechanism, we add a measure for the social severity of the crisis to our analysis asking respondents how severe they perceive the social impact of the corona crisis, measured on a scale from (1) ‘not severe at all’ to (5) ‘very severe’. We thus test whether our treatment has made respondents perceive the social consequences of the pandemic as more severe and this way affected governmental trust. The results do not provide evidence for this alternative mechanism. While respondents receiving our treatment regard the social impacts of the coronavirus pandemic as more severe (0.243, SE = 0.084, p = 0.004), this has not affected their levels of trust in the government, resulting in an insignificant mediation effect (−0.000, SE = 0.010, p = 0.964). The effects of our key mechanisms do not change.

14 Indeed, we find such an effect when using our measure of respondents’ perceptions of the government’s problem-solving capacity as a moderator of our overall treatment effect as well as the individual mediation effects. As can be seen in Figure A1 in the online appendix, we find significant interaction effects for the overall effect as well as all three mediators, highlighting that a) the effect of our treatment on trust in the government via in-group trust (panel b) is only significantly positive when respondents believe that their government is capable of solving the most pressing problems of the country and b) the overall effect of our treatment on trust in the government (panel a) as well as the effects of our treatment on trust in the government via pride in democratic institutions (panel c) and attitudes towards political elites (panel d) are only significantly negative when respondents believe that their government cannot solve the country’s most pressing problems. Naturally, this can only provide a first explanation of our results and not a robust test, given that this variable was not measured in advance of the treatment.

Additional information

Funding

This research was generously funded by the Bern University Research Foundation.

Notes on contributors

Julian Erhardt

Julian Erhardt is a PhD candidate at the Institute of Political Science, University of Bern, Switzerland. His research interests lie in the fields of public opinion research and political sociology with particular focus on political trust and alienation, support for democracy, national identity as well as social capital.

Markus Freitag

Markus Freitag is Professor of Political Sociology and of Political Psychology at the Institute of Political Science, University of Bern, Switzerland. He has published on trust, social capital, direct democracy, comparative public policy, voter participation, political attitudes, civil war, and personality.

Steffen Wamsler

Steffen Wamsler is a PhD candidate at the Institute of Political Science, University of Bern, Switzerland. His research interests lie in national identity, social integration with a focus on relative deprivation and social trust, political trust, populist attitudes, and democratic support.

Maximilian Filsinger

Maximilian Filsinger is a PhD candidate at the Institute of Political Science, University of Bern, Switzerland. His research interests lie in the fields of comparative public opinion research and political sociology. His research has been published in European Societies, German Politics and Nations & Nationalism.